Introduction

The inferior mesenteric artery (IMA) is a vessel which is usually ligated during left colon cancer surgery. However, its anatomy is rarely studied when planning a surgical approach during the preoperative patient assessment. The reason for the latter is high or routine low ligation after LCA debranching when IMA features are not relevant as long as there is no goal to preserve its branches [1]. Different variations of its branching were already described [2–5], which might be important in vessel-preserving segmental resections with D3 lymph node dissection [6, 7]. However, they are hardly used in clinical practice.

Contrast-enhanced computed tomography (CT) is a commonly used diagnostic preoperative tool for patients workup. It can be used not only for confirming the clinical diagnosis and distant metastasis, but also for defining vessel anatomy [8, 9]. Radiologists do not routinely describe the vascular anatomy of the colon. So surgeons should be able to use CT in IMA branching workup to perform anatomy-oriented segmental resections. This preoperative step is necessary to plan an approach to IMA skeletonisation if selective branches ligation is supposed.

The aim of our study was to evaluate the surgeons’ potential in accurate preoperative surgical IMA anatomy determination in patients with colorectal cancer. As a result, we also defined the most common types of IMA branching, which were found to be practically significant when D3 lymph node dissection with IMA skeletonisation and selective vascular ligation is performed.

Materials and methods

Study design

This was a prospective observational single-centre cohort study. The study protocol was approved by a local ethics committee (project No. 02-A3/2021). The informed consent was obtained from all the patients preoperatively prior to inclusion in the study.

Patients were included according to the following criteria:

— Adenocarcinoma of the splenic flexure, descending colon, sigmoid colon, rectosigmoid junction, and rectum;

— CT scans with intravenous contrast enhancement available preoperatively;

— Planned segmental resection with selective IMA branches ligation (i.e. LCA preservation in distal tumours and SRA preservation in proximal tumours).

The patients who eventually had high IMA ligation without vessel skeletonisation because of unplanned intraoperative reasons were excluded.

CT and 3D-modelling protocol

CT was performed using a Toshiba Aquillion Prime 64-detector and Siemens Somatom Sensation 40-detector scanners. The exposition parameters were defined in accordance with patient characteristics and included the following: tube potential, 120 kVp; tube current, 300—550 mA. The 0.5—1.0 mm slices were generated. Contrast agent Omnipaque (350 mgI/mL) was injected with a rate of 3.0—3.5 ml/s (1.0—1.5 ml to 1 kg of body mass). The scan timing depended on bolus tracking: the arterial phase was taken as soon as the signal within the abdominal aorta reached 150—180 Hounsfield units. The venous phase was reached 50 s after the contrast injection. The delayed phase was reached 10 min after contrast enhancement. The scans were analysed using the Osirix Dicom Viewer (Pixmeo SARL, Geneve, Swiss) or RadiAnt DICOM 2.2.9 Viewer (Medixant, Poznan, Poland). The 3D obtained images were analysed with 3D volume rendering to obtain detailed images of the vasculature.

IMA branching pattern evaluation

A group of surgeons (n=13) of a single centre was specially trained to evaluate the CT scans with intravenous enhancement to study the IMA trunk and its branches. Training included 2 educative sessions with radiologists to use DICOM Viewer software and 3D volume rendering to identify IMA branching patterns. Each surgeon further performed unaided identification of IMA anatomy using CT series of at least 20 patients, and the results were discussed with radiologists. It took them 1 month in general to finish the training program (2 days of sessions and daily CT-analysis of surgical patients’ anatomy).

The first step of the current study was CT with intravenous contrast performed preoperatively. The CT scans were studied by the surgeons, who identified the branching pattern and measured the distance from the IMA root to the aorta bifurcation and from the IMA root to the left colic artery (LCA). First, the investigator defined the pattern of LCA and its branches, i.e. the proximal part of IMA (Fig. 1). Then, the surgeon examined the IMA and its branches distally to LCA till superior rectal artery bifurcation, i.e. distal part of IMA.

Fig. 1. Scheme of proximal IMA anatomy assessment.

The first sigmoid artery arises at a distance of 0.5 cm from the left colic artery base — (a); the first sigmoid artery arises along with the left colic artery from a single point — (b).

The prospective database included the following characteristics of IMA anatomy (Fig. 1):

1) the level of LCA outflow (in mm) and its branches,

2) first sigmoid artery from LCA and/or IMA distally to LCA;

3) additional sigmoid arteries from distal segment if SRA is supposed to be preserved.

Each of those steps were schematically depicted with the use of PowerPoint software and presented to the chief surgeons (P.V.T., S.K.E. and I.A.T.) , who have more than 10-year experience in segmental colon resections with IMA skeletonisation and D3 lymph node dissection. They made final decision on IMA anatomy features, appropriate extent of skeletonisation and level of arterial ligation.

The following definitions were proposed to standardise the approach to IMA branching description:

— The LCA is the artery arising from the IMA and runs to the splenic flexure and descending colon.

— The sigmoid arteries (SAs) are vessels that go for some distance up to the sigmoid colon.

— The superior rectal artery is the continuation of the inferior mesenteric artery after the latter one gives LCA.

The IMA is divided into proximal and distal parts for interpretational and surgical purposes. The IMA trunk up to the LCA is considered as the proximal part of the IMA. As soon as the LCA debranches, the distal part of the IMA begins (Fig. 1). The accuracy of IMA branches identification was based on anatomical variations of proximal or distal parts, defined within the extent of skeletonisation.

Two types of IMA proximal part were identified:

a) An independent arterial trunk (AT) with a length of more than 0.5 cm. The length of 0.5 cm is optimal enough from a surgical perspective as the trunk can be cut at this distance by two clips and electrosurgical instrument. In our cases, the length of the segment is defined after sharp skeletonisation intraoperatively by 2 clips put in this segment for further transection with ultrasound scalpel.

b) A fan-shaped branching of LCA and first SA from one point of the IMA trunk or common trunk, giving its branches less than 0.5 cm from IMA.

The next step was evaluation of the IMA anatomy and its branches distally and spatial relationship of colic vessels, i.e. if LCA is above or below inferior mesenteric vein.

The surgeons depicted the LCA trunk on diagram 1 and all its branches both in proximal and distal parts preoperatively to demonstrate the features of its anatomy found with the use of CT and 3D-CT in some cases (Fig. 2).

Fig. 2. CT-based vascular anatomy and intraoperative data.

Preoperative scheme of vascular anatomy — (a); 3D CT angiography showing the structure of inferior mesenteric artery — (b); intraoperative assessment of proximal IMA during skeletonization (intraoperative image) — (c); scheme showing intraoperative data (distal sigmoid artery identified in specimen) — (d).

IMA branching and the positional relationship of the vessels were approved intraoperatively by IMA and LCA or SRA skeletonisation. Intraoperatively obtained data was documented with photo or video approval (Fig. 2). The diagram 2 was constructed according to the intraoperative photo (Fig. 1).

The surgeon’s role in CT-based anatomy identification was evaluated by the concordance of preoperatively found IMA branching pattern within the extent of skeletonisation (Fig. 2). If distal resection margin of sigmoid colon was above promontorium , the branching of distal IMA was examined as well. If all characteristics that were available to determine by skeletonisation were proved intraoperatively, preoperative IMA anatomy assessment was considered right.

The results obtained by surgeons preoperatively and intraoperatively, were introduced into the database. Finally, preoperative and intraoperative diagrams of IMA were compared. Intraclass correlation between diagrams 1 and 2 i.e. types of LCA and/or SRA branching were assessed.

The primary endpoint of the study was to evaluate if surgeons were accurate enough in identifying the LCA and SRA branches. The secondary endpoint was to delineate surgically relevant IMA branching pattern i.e. for purposes of vessel preservation with D3 lymph node dissection. The latter is available for cases with skeletonisation of distal part of IMA with SRA preservation.

Surgical approach

The segment of the bowel with the tumour was resected with preservation of the visceral fascia and D3 lymph node dissection up to the root of the inferior mesenteric artery. The branches supplying the tumour were selectively ligated. To perform selective arterial ligation with D3-LND, the IMA trunk and its branches were skeletonised.

The skeletonisation margins differed depending on tumor location and volume of bowel resection:

— If the tumour was in splenic flexure, IMA was skeletonised 1 cm distally to LCA outflow — the latter was ligated, splenic flexure resection was performed. LCA was skeletonised till sigmoid artery debranching.

— If the tumour was in the sigmoid colon distally (i.e. the distal resection margin was less than 10 cm from promontorium), rectosigmoid junction or rectum, IMA skeletonised approximately 1 cm distally to LCA outflow. LCA was skeletonised till SA debranchingThe radiologists routinely used “sigmoid take-off” as main anatomical landmark to define if tumor was in distal sigmoid colon, rectosigmoid junction or rectum (10). Thus, the correctness of surgeons’ decisions in these locations was evaluated based on proximal IMA interpretation.

— If the tumour was in descending and sigmoid colon proximally, IMA and SRA were skeletonised until the mesocolon distal resection margin (10). Left colon segmental resection or sigmoid colon resection was performed with SRA preservation. Surgeon’s preoperative evaluation was examined for proximal and distal part of IMA in these cases.

The IMA trunk was dissected from mesocolic tissue with lymph nodes from group 253. All patients had complete mesocolic excision (CME) with bowel resection margins of 10 cm proximally and distally from main feeding vessel. If the tumour was located in the rectum, a 3 cm distal resection margin was considered according to the Japanese guidelines for colorectal cancer treatment (11). The main anatomical landmark in this area was the proximal part of the IMA. Then, a skeletonisation of the distal IMA segment was performed if SAs and/or LCA were supposed to be ligated. That approach provided an opportunity to preserve blood flow in the IMA to remnant segments of the large bowel and perform complete lymph node dissection at the same time.

Interpretation

The data was considered as concordant if the preoperative interpretation was the same intraoperatively within skeletonisation margins, as the main goal of the study was to define the surgeon’s role in vessel anatomy interpretation for D3 lymph node dissection with vessel preservation.

Statistical analysis

The concordance of preoperative scheme with intraoperatively defined branching pattern was evaluated statistically by intraclass correlation analysis to find the concurrence between observers’ CT-based anatomy assessment and intraoperative data (diagram 1 vs. diagram 2).

The sensitivity and specificity, which is characterized by co-agreement of the results, obtained by CT and intraoperatively, was proved by Kendall’s tau-b test, where 1 stands for perfect agreement.

Intra-reliability and Kendall’s tau-b were performed both for IMA type and positional relationship of IMA and IMV.

The IBM SPSS statistics version 23 software package was used for statistical analysis and calculation of the sample size and graphics.

Results

The features of IMA anatomy were described preoperatively in patients treated for left colon or rectal cancer. As a result, 250 patients with segmental colon resection and selective IMA branches ligation from January 2018 to March 2020.

Patient characteristics are presented in Table 1.

Table 1. Patient characteristics

The analysis of IMA anatomy based on routine CT imaging showed significant variability in LCA and SAs branching patterns. As a result of preoperative analysis of IMA branching, we identified that IMA features can be described by the branching of LCA and SA in the proximal and distal parts. If colic vessels were debranching in both the proximal and distal parts, there were more opportunities for the surgeons to preserve blood flow in both the caudal and cranial directions. We analysed the branching pattern of SAs in distal part if SRA was preserved and demonstrated the percentages of their occurrence (Table 2). The number of sigmoid arteries varied from 1 to 3 branches in distal IMA in our cohort of patients.

Table 2. Inferior mesenteric artery branching patterns

As a result of the IMA evaluation, the branching pattern can be associated with the letters E, K, and H (Fig. 3) to standatise the interpretation. If there are one or more sigmoid arteries in distal part of IMA, the scheme depicting IMA anatomy can be associated with E-letter. We identified three subtypes of E-shaped IMAs when proximal and distal IMA were skeletonised.

Fig. 3. IMA branching patterns and 3D reconstruction.

In the E1 type, several colon branches independently start from the IMA trunk (Fig. 3) .

In the E2 type LCA and SA start from the IMA at one point in a fan-shaped fashion in the proximal part, while SA can be also found in the distal part (Fig. 3) .

In the E3 type, LCA starts as an independent branch, and SAs debranch from SRA in a fan-shaped mode (Fig. 3).

The K type is considered if LCA and SA start from a point or root less than 0.5 cm in length in the proximal part of the IMA if there were no additional sigmoid branches distally.

H-type is supposed if LCA and SA start from common root longer than 0,5 cm, which is described as colosigmoid trunk elsewhere(12), and SRA does not give hemodynamically significant branches (i.e. those not seen by contrast enhancement) (Fig. 3).

Intraclass correlation showed a high level of agreement between preoperative 3D-CT modelling by surgeons with an intraoperative approach to evaluate features of IMA anatomy IC=0.926; CI=0,906-0,941; p < 0.001).

The correlation coefficient (CC) was assessed with Kendall’s tau-b test and showed high concordance of the results obtained with CT and compared with intraoperative data (CC=0.912; p <0.001).

The E1 type was the most common type in our cohort of patients. The prevalence of proximal and distal IMA branching patterns is shown in Table 2.

Surgeons were correct in IMA anatomy evaluation in 94,8% (237 out of 250) of cases. The inferior mesenteric vein (IMV) was dorsally to the LCA in 83,8% of cases. Concordance in IMV spatial determination was measured with intraclass correlation (IC=0,971; CI = 0,963-0,977). The IMV position relatively to LCA was determined in a right way in 96,9% of cases.

Statistical analysis showed high concordance between the results obtained with preoperative CT and intraoperative identification.

Surgical interpretation resulted in a discordance in 13 cases (5,2%), which were analysed by CT-specialist.

Discussion

The present study is the result of daily surgical practice in our centre, where a personalised approach to surgical treatment based on the anatomical features of IMA is used in consideration with tumour location. Routine CT with intravenous contrast provides an opportunity not only to stage the tumour, but also to determine the anatomical variation of the feeding vessels. However, the standard CT protocol introduced by radiologists does not commonly describe IMA anatomy according to existing classifications.(1-4). They are not routinely used by surgeons in case of segmental resection and skeletonisation planning. However, the ignorance of IMA anatomy may have a negative impact on surgical outcomes, especially if D3 lymph node dissection with vessel preservation is supposed. All classifications already described in the literature are based on CT or cadaveric dissection findings. To our knowledge, there have been no attempts to identify IMA branching types according to intraoperative findings and their correlation with preoperatively obtained data. This comparison shows the admissibility of IMA anatomy examination by surgeons with the use of CT. A low percentage of mistakes made during preoperative evaluation of IMA anatomy by surgeons showed high accuracy of CT in purpose of vessel preservation and D3 lymph node dissection.

The role of the surgical team in CT-based IMA anatomy interpretation for surgical purposes by intraoperative verification has not yet been described. To the best of our knowledge, this approach has not been previously used and implemented because of the technical aspects of CME for left-sided cancer, which supports high vascular ligation of IMA (13). Nesgaard et al. used the same approach in the superior mesenteric artery (SMA) anatomy study, which is supported by SMA skeletonisation for surgical purposes if D3 lymph node dissection is performed (6).

In our study, we have demonstrated the availability of preoperative planning with the use of CT and 3D reconstruction to determine IMA anatomy in routine surgical practice.

However, IMA anatomy was mistakenly assessed by surgeons in 5,2% of cases.

CT-specialists analysis of CT-scans and surgeons responds came to the following conclusions:

— in 3 cases SAs were not found on CT debranching from distal IMA, when SRA was supposed to be skeletonised. Additionally, they are hardly found if there is a redistribution of blood flow into dominant artery, supplying the most of left colon;

— in 6 cases sigmoid branches debranching from LCA were not found. The reason for the latter was the hypocontrasting of the sigmoid branches with small calibre and associated defects of visualisation.

— in 3 cases surgeons identified mesenteric branches as SAs, debranching from LCA or at the same point from IMA. In these cases 3D-CT is a helpful tool for branches differentiation.

Thus, E subtypes require more contrasting agents, higher speed of infusion, and vasodilators. However, this approach is not routinely used in CT for clinical diagnosis and metastatic lesion exclusion.

IMV position was mistakenly identified in 2,6% of cases. In analysis, performed by CT-specialist it was determined, that in 4 cases IMA was behind sigmoid branch, but in front of LCA.

Surgeons usually do not investigate IMA distal to LCA debranching, because they are routinely ligated at this level for anterior resection of the rectum. The level of LCA outflow is the main anatomical landmark in lymph node dissection at both D2- and D3-level. The IMA study approach introduced in this study may help to standardise vessel anatomy identification.

The suggested rubrication of IMA anatomy is based on surgical needs of skeletonisation. The level of LCA branching is important in low ligation for distal sigmoid and rectal cancer. Identifying colic branches, starting from the same level with LCA or distally from the latter one, is important to perform LCA skeletonisation and D3 lymph node dissection at its root and be aware of their outflow level.

Preoperative evaluation of IMA anatomy and obtained data are used to plan skeletonisation of vessels in segmental resections with vascular preservation. Previously, it was demonstrated that central vascular ligation can result in a decreased postoperative quality of life and survival rates(14,15) . Akagi et al. showed an increase in overall survival in D3-lymph node dissection patients with left colic artery preservation with sigmoid and rectal cancer (16). Thus, preservation of the sigmoid and superior rectal arteries can result in the same outcomes for splenic flexure and descending colon cancer.

One of the limitations of the study is the impossibility to examine the distal IMA fully, if SRA is not skeletonised. However, we recently started to study IMA distally to LCA, if IMA was cut right after LCA. Their clinical significance is yet to be determined (12).

Merkel et al. demonstrated that CME implementation in surgical practice as a standard approach decreased the frequency of local recurrences and distant metastases and increased overall and cancer-specific survival (17). However, one of the crucial steps of this surgical procedure is hith IMA ligation at the root. It has been suggested that IMA ligation at its origin allows performing of extended lymph node dissection in CME (18–20). At the same time, high vascular ligation at the IMA root can result in ischemia at the line of anastomosis and preserved part of the left colon(21,22). The preservation of blood flow via the IMA and its branches which are not involved in the oncological process with excision of lymph nodes along feeding vessels is achievable due to the skeletonisation of the IMA and its proximal part. So, we considers complete mesocolectomy and D3-lymph node dissection may be performed with selective vascular ligation if the anatomy of the main feeding vessel is determined.

In left colon cancer, it is important to perform an organ-preserving approach with selective vascular tie instead of standard hemicolectomy and extended resections, as they are associated with a higher quality of life and survival rates(23,24). The risk of tumour spread in the proximal and distal directions above 10 cm from the macroscopically identified tumour is significantly low. These data provide evidence for Japanese guidelines, which suggest an approach to surgical margins in colon tumour resections with oncological principles (8). These principles may be followed by selective ligation of the IMA branches if its anatomy is comprehensively understood.

Conclusion

CT data provides an opportunity to define the IMA anatomy with high accuracy. CT results obtained by surgeons do not differ significantly from the radiologists’ conclusions. Thus, routine CT-based preoperative identification of the IMA anatomy by surgeons may be recommended in clinical practice.

The E-, K-, and H- IMA branching types could be practically significant and may be used in routine practice for preoperative planning of surgical tactics.

The authors declare no conflicts of interest.